| Message ID | 20250319152927.1263033-1-meetsoni3017@gmail.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | reftable: return proper error codes from block_writer_add | expand |
On Wed, Mar 19, 2025 at 08:59:24PM +0530, Meet Soni wrote: > This patch series attempts to avoid making an assumption regarding error codes > returned by block_writer_add(). > > Changes since v4: > - update commit message. > - add documentation comment. One additional change that isn't mentioned here is that we now check for REFTABLE_ENTRY_TOO_BIG_ERROR the second time we call `block_writer_add()` when writing object records, which is what my only concern was. So with that now addressed I'm happy with this patch series, thanks for working on it! Patrick
This patch series attempts to avoid making an assumption regarding error codes returned by block_writer_add(). Changes since v4: - update commit message. - add documentation comment. Meet Soni (3): reftable: propagate specific error codes in block_writer_add() reftable: adapt writer_add_record() to propagate block_writer_add() errors reftable: adapt write_object_record() to propagate block_writer_add() errors reftable/block.c | 13 ++++++------ reftable/block.h | 2 +- reftable/record.c | 53 +++++++++++++++++++++-------------------------- reftable/writer.c | 34 +++++++++++++++++++++--------- 4 files changed, 56 insertions(+), 46 deletions(-) Range-diff against v4: 1: 6ab35d569c = 1: 6ab35d569c reftable: propagate specific error codes in block_writer_add() 2: 7f0bdc27e1 ! 2: 873a991a2c reftable: adapt writer_add_record() to propagate block_writer_add() errors @@ Metadata ## Commit message ## reftable: adapt writer_add_record() to propagate block_writer_add() errors - Previously, writer_add_record() would flush the current block and - retry appending the record whenever block_writer_add() returned any - nonzero error. This forced an assumption that every failure meant - the block was full, even when errors such as memory allocation or I/O - failures occurred. + Previously, writer_add_record() would flush the current block and retry + appending the record whenever block_writer_add() returned any nonzero + error. This forced an assumption that every failure meant the block was + full, even when errors such as memory allocation or I/O failures occurred. - Update the writer_add_record() to inspect the error code returned by - block_writer_add() and only flush and reinitialize the writer when the - error is REFTABLE_ENTRY_TOO_BIG_ERROR. For any other error, immediately - propagate it. + Update the writer_add_record() to inspect the error code returned by + block_writer_add() and only flush and reinitialize the writer when the + error is REFTABLE_ENTRY_TOO_BIG_ERROR. For any other error, immediately + propagate it. Signed-off-by: Meet Soni <meetsoni3017@gmail.com> 3: 480ac27797 ! 3: 1e2f7ff83f reftable: adapt write_object_record() to propagate block_writer_add() errors @@ Metadata ## Commit message ## reftable: adapt write_object_record() to propagate block_writer_add() errors - Previously, write_object_record() would flush the current block and - retry appending the record whenever block_writer_add() returned any - nonzero error. This forced an assumption that every failure meant the - block was full, even when errors such as memory allocation or I/O - failures occurred. + Previously, write_object_record() would flush the current block and retry + appending the record whenever block_writer_add() returned any nonzero + error. This forced an assumption that every failure meant the block was + full, even when errors such as memory allocation or I/O failures occurred. - Update the write_object_record() to inspect the error code returned by - block_writer_add() and only flush and reinitialize the writer when the - error is REFTABLE_ENTRY_TOO_BIG_ERROR. For any other error, immediately - propagate it. + Update the write_object_record() to inspect the error code returned by + block_writer_add() and flush and reinitialize the writer iff the + error is REFTABLE_ENTRY_TOO_BIG_ERROR. For any other error, immediately + propagate it. - All call sites now handle various error codes returned by - block_writer_add(). + If the flush and reinitialization still fail with + REFTABLE_ENTRY_TOO_BIG_ERROR, reset the record's offset length to zero + before a final attempt. + + All call sites now handle various error codes returned by + block_writer_add(). Signed-off-by: Meet Soni <meetsoni3017@gmail.com> @@ reftable/writer.c: static void write_object_record(void *void_arg, void *key) arg->err = writer_flush_block(arg->w); if (arg->err < 0) goto done; +@@ reftable/writer.c: static void write_object_record(void *void_arg, void *key) + if (arg->err < 0) + goto done; + ++ /* ++ * If this still fails then we may need to reset record's offset ++ * length to reduce the data size to be written. ++ */ + arg->err = block_writer_add(arg->w->block_writer, &rec); + if (arg->err == 0) + goto done; + ++ if (arg->err != REFTABLE_ENTRY_TOO_BIG_ERROR) ++ goto done; ++ + rec.u.obj.offset_len = 0; + arg->err = block_writer_add(arg->w->block_writer, &rec); +